A Single Amino Acid Polymorphism Within the α2 Domain of HLA-DQα Regulates the Assembly and Expression of HLA DQ Alleles and the Potential Clinical Impact

Abstract

Background.

Single antigen bead testing is currently accepted as standard practice for evaluating HLA alloantibodies. However, it is well established that the expression of HLA antigens, on beads, is not representative of actual cellular expression. It is well recognized that the cellular surface expression of certain HLA loci are expressed at relatively low levels. As such, single antigen bead data may overrepresent the impact of HLA antibodies for certain loci. Importantly, for clinical assessments, the cell surface expression of target donor HLA antigens is not routinely considered unless it is a known null allele.

Methods.

Using a panel of variably reactive DQ4 allosera, we have evaluated DQ4 antigen assembly and surface expression on cell lines and in transfectants carrying DQ4 variants.

Results.

A single amino acid change at residue 137 in the α2 domain of the DQα chain can significantly affect the stability of the αβ heterodimer and result in the reduced surface expression of the affected DQ antigens, resulting in changes in anti-HLA antibody reactivity.

Conclusions.

A single amino acid residue in the DQα chain can influence the expression of a DQ4 molecule. Importantly, such surface expression variability does significantly impact allo-antibody reactivity. As a result, amino acid sequence variations in the α2 and β2 domains of the HLA class II serological groups and alleles within serological groups should be further investigated to understand the scope of allelic expression variability. Identifying low surface expressing HLA antigens/alleles might facilitate the understanding of certain discordant clinical observations.

INTRODUCTION

An HLA class II molecule is a heterodimer composed of an α and β polypeptide chain, each containing 2 distinct extracellular domains. The β chain for HLA-DR, -DQ, and -DP is highly polymorphic, whereas the α chain for DR is dominated by 1 common allele. In contrast, both DQα and DPα chains are highly polymorphic. The α1 and β1 domains create the peptide-binding groove located at the molecule’s apex. This groove specifically binds endogenous or exogenous peptides to be presented by antigen-presenting cells to CD4⁺ T cells, thereby initiating an immune response.¹

Most of the polymorphic amino acids of the HLA class II molecules are located on the α1 and β1 domains and can alter the structure within and around the groove thus influencing the peptide repertoire for each class II heterodimer. Because peptide repertoire directly affects the cellular immune response, algorithms based on the amino acid sequences of HLA class II α1 and β1 domains have been developed to predict the peptide repertoire capable of binding to a unique α and β allele pair. Such predictions are then used for assessing the interaction between antigen-presenting cells and T-cell receptors. Importantly, how polymorphisms within the α2 and β2 domains may impact assembly and expression of the heterodimers have been lacking.^2,3

It has been shown that cis-associated DQα and DQβ chain combinations co-evolved to form favorable heterodimers. Trans-associated DQα and DQβ chain combinations, although real, exist at a greatly reduced frequency. The ability of α and β chains to form stable heterodimers can be separated into 2 categories based on serological groupings: DQ1 (DQA1*01 with DQB1*05/06) and DQ2/3/4 (DQA1*02/03/04/05/06 with DQB1*02/03/04).^4-6 It was also shown that amino acids 60–91 of certain DQB1 alleles might be the barrier for formation of intergroup heterodimers.⁵ Even though there were attempts to elucidate the critical amino acids in the DQ α/β heterodimer assembly by comparison of the recombinant variant protein expression on the murine and insect cell surface, it was limited to the polymorphic positions in the α1 and β1 domains,⁶ The potential impact of polymorphic amino acid residues on the α2 domain of DQA1 alleles to affect αβ heterodimer formation has not been reported.

In our recent study, we observed weak antibody reactivity of some DQ4-specific sera to the DQB1 antigen that was associated with DQα0402 but not with DQα0401.⁷ Because there is only 1 amino acid difference at position 137 between these 2 DQα alleles, we hypothesized that the amino acid at position 137 might affect the assembly and surface expression of these αβ heterodimers. In this study, we first identified sera that can distinguish between the DQ4 α and β chains and used them to examine the expression level of both chains using purified DQ antigens from B-cell lines. These experiments were designed to confirm the differential DQ4 reactivity observed on phenotype antigen panel. Additionally, using transfectants, we sought to explore the effect of amino acid changes at amino acid position 137 of the DQα chain in different serological groups. Our findings demonstrated that amino acid residue 137I in the α2 domain of DQα0402 not only impairs heterodimer assembly and expression when paired with DQβ0402 but also when paired with other DQβ02 or DQβ03 alleles. Furthermore, amino acid changes in residue 137 of other DQα alleles can impact heterodimer assembly and surface expression when combined with their specific DQβ alleles.

MATERIALS AND METHODS

Cell Lines and Antibodies

Detail information of Epstein-Barr virus (EBV)-transformed B-cell lines used in this study can be found in the Supplementary Materials and Methods (SDC, https://links.lww.com/TP/D447) and in Table S1 (SDC, https://links.lww.com/TP/D447). Monoclonal antibodies FM5148 (anti-DQα), FS1754 (anti-DQβ04) and T2DM, a T2 cell line transfected with HLA-DM (DMA*01:01 and DMB*01:01) cDNA, were gifts from Thermo Fisher Scientific, Transplant Diagnostic Division. T2DM has no detectable expression of HLA class II on the cell surface.

HLA Antibody Detection Assay

HLA class II antigens were purified through cell lysis and affinity purification from cell pellets as described.⁸ A pan HLA-DQα-specific antibody, FM5148 (Figure S1, SDC, https://links.lww.com/TP/D447), was used to isolate HLA-DQ antigens from both heterozygous (hetDQ) and homozygous (monoDQ) EBV-transformed B-cell lines or transfectants as previously reported.⁷ The purified antigens were then coupled to Luminex beads, and a solid-phase bead assay was performed according to the LABScreen antibody detection protocol.⁹ Results were analyzed using a LABScan3D flow analyzer (Thermo Fisher Scientific, West Hills, CA). Commercially available HLA class II antibody detection kits (LS2PRA and LS2A01) were obtained from Thermo Fisher Scientific, Transplant Diagnostic Division.

Patient Serum Screening

Since many of the DQ4 antigens used in this study were derived from heterozygous B cell lines, they are likely derived from the DQA1*04-DQB1*04 haplotype. To ensure the selected DQ4 positive sera would be reactive to the DQ4 antigen in the heterozygous cell lines, we selected sera that have at least 3000 mean fluorescent intensity (MFI) to the DQ4 single antigen beads (SABs). Eighty-one (81) DQ4 positive serum samples (MFI ≥ 3000) were identified from 878 patients awaiting or post kidney transplants by analyzing data from the HLA class II single antigen bead panel (LS2A01, One Lambda). Since most of the DQ4-positive sera have additional reactivity to other DQ antigens, these 81 DQ4-positive sera were further tested with a panel of DQ antigens purified from homozygous B-cell lines and an HLA class II phenotype bead panel (LS2PRA, One Lambda) to define the detailed DQ antibody specificity for each serum.⁷ DQα04-specific sera were defined as reacting to DQα04 heterodimers that are paired with both DQβ04 and non-DQβ04 alleles. DQβ04-specific sera were those that reacting to DQβ04 alleles paired with both DQα04 and non-DQα04 alleles. Sera were excluded if they have antibody reactive to both DQα04 and DQβ04 chains (n = 19). To avoid the interference of the DR or DP antibody in sera, only purified DQ antigens were used in this study (Table S2, SDC, https://links.lww.com/TP/D447). Because DQ antigens purified from heterozygous cell lines were used in this study, only those sera that have reactivity <3000 MFI to the DQ SABs representing the second DQ antigen were selected to minimize the interference of extra non-DQ4 antibody reactivity. Qualified sera were separated into either DQ4 α chain or DQ4 β chain specific group based on the typing of the DQ antigen panels and were identified as depicted in Figure 1.

FIGURE 1.

Serum screening and categorizing strategies. Samples were screened by commercially available kits and then confirmed by the monoDQ antigens.⁷ Samples were categorized based on their specific HLA-DQ specificities and grouped for the DQα0402 vs DQα0401 studies.

The use of human sera was approved and written consent from patients were waived by the Methodist Specialty and Transplant Institutional Review Board (IRB) Board of Directors, Emory University IRB, Mass General Brigham IRB (Protocol #2025P002064), Cantonal Ethics Committee in Zurich, or were leftover and deidentified sera provided by Texas Medical Specialty laboratory.

Transfection, Cell Culture, and Flow Cytometry Analysis

Constructs for the site-mutagenesis on DQA1 cDNA used for this study and generating stable cell lines via Neon Transfection System (Thermo Fisher Scientific, Waltham, MA) can be found in the Supplementary Materials and Methods (SDC, https://links.lww.com/TP/D447). A R-phycoerythrin conjugated streptavidin (SAPE) enhanced staining protocol for flow cytometry was used to detect DQ on the EBV-transformed B cells with human serum.¹⁰

HLA Protein Databank 3D Model and Statistical Analysis

AlphaFold (https://alphafoldserver.com/) was used to predict and generate DQβ0402:DQα0401 PDB files. Mol* 3D Viewer (https://www.rcsb.org/3d-view/) was used to perform the 3D analysis. ANOVA was used to perform statistical analysis.

RESULTS

DQα0402 Is Responsible for the Low DQ4 Antigen Surface Expression in DQB1*04:02-DQA1*04:02 Haplotype

It has been observed that many serum samples assigned as DQ4-positive based on the phenotype antigen panel (LS2PRA, One Lambda) have low or negative reactivity to the DQ4 antigen extracted from E20781 B-cell line (DQB1*04:02-DQA1*04:02), whereas the other 6 DQ4 antigen bearing beads (DQB1*04:02-DQA1*04:01) have much higher reactivity (Figure 2A). In contrast, some other DQ4-positive sera show strong reactivity with E20781 (Figure 2B). To further investigate this phenomenon, purified DQ antigens from the same 7 heterozygous B-cell lines that were used to produce the LS2PRA panel were tested with 2 groups of DQ4-reactive sera (Figure 1) to verify the observation on phenotype antigen panel: group 1 (n = 4) sera reactive to the β chain of the DQ4 heterodimer; and group 2 (n = 19) sera reactive to the α chain of the DQ4 heterodimer. All serum samples that were reactive to DQα04 also reacted to DQα05 and 06 alleles (Figure S2B, SDC, https://links.lww.com/TP/D447) and most likely recognize the Epitope (eplet) #2018 identified for DQα04/05/06,¹¹ which is shared by both DQA1*04:01 and DQA1*04:02 alleles; therefore, both alleles should be reactive to the DQα04/05/06 antibodies. The serum samples that were reactive to DQβ04 chain (n = 4) showed similar reactivity to all 7 DQ4 antigen preparations including E20781 (P > 0.4, Figure 2C). In contrast, serum samples that were reactive to DQα04 (DQα04/05/06, n = 19) showed a significant reduction in reactivity to the DQ4 antigen purified from E20781 (DQβ0402:DQα0402) (P < 0.00001, Figure 2D). As the beads used were heterozygous for DQ, to account for any secondary or cross-reactivity to the second DQ antigen, a panel of individual DQ alleles representing all 5 second antigens were created. Both groups of sera demonstrated negative reactivity to the DQ antigens purified from the homozygous cell lines expressing only the second DQ antigens representing the 7 DQ4 heterozygous cells. These studies confirmed that the sera selected for study were monospecific for DQ4 and did not possess any additional polyclonal antibody reactivity (Figure 2E).

FIGURE 2.

DQ4 reactive sera exhibited distinct reaction patterns against DQ4 bearing antigens from a phenotype antigen panel (LS2PRA, One Lambda). A, Reaction pattern of an E20781-negative DQ4 serum on the LS2PRA panel. B, Reaction pattern an E20781 positive DQ4 serum on the LS2PRA panel. DQ4 antigens were boxed in magenta and E20781 is highlighted and boxed in dark blue. C, Reaction profile of DQβ04 positive samples on the DQ antigens from the DQ4-positive cell lines. D, Reaction profile of DQα04/05/06 positive samples on the DQ antigens purified from the DQ4-positive cell lines. E, All sera tested showed negative reactivity to the non-DQ4 DQ antigens on the heterozygous DQ4 antigen panel. Because the second allele for the 7 DQ4 heterozygous cells was either DQ0501, DQ0503, DQ8 or DQ9, the DQ4 reactive serum samples (n = 23) that were selected for study were all negative to DQ0501, DQ0503, DQ8, and DQ9 to eliminate the interference of the extra reactivity from the second DQ antigen. Blue lines are the threshold for pos/neg reactivity used for this study.

The DQβ0402 peptide chain in E20781 was detected at similar levels as the DQβ0402 antigen from the other 6 cell lines (Figure 2C), indicating that the expression of DQβ0402 in E20781 results from efficient heterodimer formation with the DQα0301 which is in trans, and confirmed by the robust reactivity of DQα-specific monoclonal antibody FM5148 (Figure 2D). The observed allelic variation in reactivity to DQα04/05/06 antibody suggests that either poor expression of DQα0402 itself or low efficiency of the expressed DQα0402 polypeptide to form an αβ heterodimer with DQβ0402 could be responsible for the weak antibody reactivity to E20781 extracted antigens.

Low DQα0402 Antibody Reactivity was Confirmed in Another DQA1*04:02-positive B-cell Line

To rule out a possible genetic defect in E20781 that results in the low expression of DQα0402, DQ antigen purified from another DQA1*04:02-DQB1*04:02 cell line, E6452, was tested by the same two groups of serum samples. DQ antigens from 2 other cell lines, TER351 and TER416, which share the same DQ alleles with E20781 and E6452, respectively, except they have DQA1*04:01 instead of DQA1*04:02, were used as references for the antibody reactive levels (Table S1, SDC, https://links.lww.com/TP/D447). To eliminate the interference of reaction to the second DQ antigen, DQ8- or DQ2-negative serum samples were selected for either pair of E20781/TER351 (group A) or E6452/TER416 (group B), respectively (Figure 1). Similar reaction patterns were observed between these 2 pairs of DQ antigens: no significant difference in reactivity to DQβ04-positive sera with their DQA1*04:01 counterpart (P = 0.58528 in E20781 versus TER351 and P = 0.29338 in E6452 versus TER416, Figure 3A and D). However, a sharp decline in reactivity to the DQα04/05/06 positive sera was observed when compared with their DQA1*04:01 counterpart (P < 0.00001 in E20781 versus TER351 and P = 0.00004 in E6452 versus, TER416, Figure 3B and E). Again, both groups of sera were shown to be DQ4 specific and did not contain additional polyclonal antibody reactivity to the second DQ antigen (Figure 3C and F). Like E20781, DQβ0402 peptide chain in E6452 was detected at the similar level as the reference antigen from TER416 (P > 0.29) indicating it could form heterodimer with the DQα0201 in trans location with equivalent efficiency (Figure 3D).

FIGURE 3.

Confirmation of positive DQ4 reaction patterns in the DQ4 serum samples. A-C. E20781. D-F. E6452. G. Flow cytometry of SSA1 serum and mouse anti DQβ04 (FS1754) with three DQ4 harboring cells. Blue shaded peaks are the HLA negative controls and crimson lines are the SSA1 reactivity. Gray Shaded peaks are mouse IgM isotype controls and olive lines are Mouse anti DQβ04 (FS1754).

TER351, E20781, and IHW9068 cell lines were further tested by flow cytometry for the surface expression of the DQβ0402:DQα04 protein with SSA1 (a strong DQα04/05/06 reactive serum with low reactivity against DR and DP antigens) to confirm the reactivities observed on the SAB assay (Figure 3G). TER351 (DQα0401) showed an increased reactivity to SSA1 while E20781 (DQα0402) showed undetectable SSA1 signal over the negative control serum. The DQβ0402:DQα0401 homozygous cell line, IHW9068, exhibits the strongest positive signal. All 3 cell lines reacted positively with a DQβ04 specific monoclonal antibody, FS1754. This result confirmed the low surface expression of DQβ0402:DQα0402 heterodimer was due to the presence of DQA1*04:02 allele in E20781, and the positive reactivity to FS1754 came from the DQβ0402:DQα0301 heterodimer on the cell surface.

Inefficient Assembly With DQβ Chain and not Poor Synthesis of DQα0402 Chain Is Responsible for the low Surface Expression

To investigate of the cause of low DQα0402 expression in cells, we examined the cell surface expression of DQ heterodimers by co-transfection of either DQA1*04:01 or DQA1*04:02 allele with DQB1 alleles from DQ2, 3, and 4 serological groups. Significantly lower DQ surface expression was observed on all 8 DQA1*04:02 transfectants when compared with the corresponding DQA1*04:01 transfectants (Figure 4A). Because DQA1*04:02 expression was directed by a strong constitutive expressed promoter (EF1α) and not its native promoter, it suggests that inefficient assembly of DQα0402 with DQβ is responsible for the low abundance of DQα0402 in the DQA1*04:02-DQB1*04:02 B cell lines and not because of genetic defects that result in poor synthesis of DQα0402. Furthermore, even though DQβ0402:DQα0402 was expressed at a lower level than DQβ0402:DQα0401 on the T2DM transfectants (Figure 4B), the purified DQβ0402:DQα0402 antigen was equally reactive as the purified DQβ0402:DQα0401 antigen to all the DQβ04 and DQα04/05/06 reactive sera (Figure S2, SDC, https://links.lww.com/TP/D447). These results support the conclusion that the T137I change affects the αβ heterodimer assembly but not the antigenicity of the tertiary structure of the assembled heterodimer. It also confirms our assumption that DQα04/05/06-positive sera recognize both DQβ0402:DQα0401 and DQβ0402:DQα0402.

FIGURE 4.

Threonine in amino acid #137 (in DQα04/05/06) or AA#138 (in DQα03) affects cellular DQ expression. A, Flow analysis on the DQ expression of transfectants carrying same DQβ chain with either DQα0401 (Threonine, crimson line) or DQα0402 (Isoleucine, olive line). Shaded blue is the host (T2DM). B, Statistical analysis of the site-mutagenesis on Threonine in amino acid #137/#138 to Isoleucine on the DQα polypeptide with the same DQαβ combination.

The only difference between DQA1*04:01 and DQA1*04:02 sequence is a single nucleotide C→T change at codon 137 resulting in an amino acid change from Threonine (Thr), a polar amino acid, to Isoleucine (Ile), a nonpolar amino acid (Figure S3A, SDC, https://links.lww.com/TP/D447). To explore if α137I could affect the expression of DQ αβ heterodimer in other serological groups, we performed site-mutagenesis at codon 137 (codon 138 in DQA1*01 and *03) and changed Threonine to Isoleucine codon in DQA1*01:02, DQA1*02:01, DQA1*03:01, DQA1*05:01, and DQA1*06:01. The resulting mutant alleles were then paired with the respective DQβ chain from common haplotypes to create stable DQ transfectants. Cell surface expression of each DQ αβ heterodimer was negatively affected at various degrees by the Threonine to Isoleucine change in all DQA1 alleles (P < 0.02) with the greatest impact observed in DQB1*04:01 and DQB1*04:02 alleles (Figure 4B). It is interesting to note that a single amino acid substitution (R23L, Figure S3B, SDC, https://links.lww.com/TP/D447) in DQB1*04:01 allele lower the DQ4 αβ heterodimer surface expression even further than its DQB1*04:02 counterpart (P = 0.00067, Figure 4B).

Additional Hydrogen Bonds Between α137T Side Chain and α150I Might Contribute to the Stability of DQ4 Heterodimer

Because the location of amino acid residue 137 in DQα chain can potentially interact with other amino acids in proximity based on the existing x-ray crystal structure of multiple DQ heterodimers,^12,13 it is plausible that amino acid difference at position 137 can affect the assembling and expression via space repulsion/structure changes. To address this possibility, a series of codon 137 variants of DQA1*04:01 were constructed via site-directed mutagenesis, each with 1 of the 20 amino acids. It appears that amino acid Serine (S) and Threonine (T) with small molecular size and hydroxy bearing polar side chain are preferred in DQα:DQβ complex (Figure 5A). All other 18 amino acid substitutions result in varying degrees of reduction in DQ surface expression indicating the naturally evolved Threonine is the optimal amino acid in the assembly of the DQ αβ heterodimer. Detailed analysis on the amino acid changes in position 137 are demonstrated in Figure 5B–D and in Figure S4 (SDC, https://links.lww.com/TP/D447) using a 3D prediction software, AlphaFold 3.0 (https://alphafoldserver.com/) to generate the structures and analyzed by Mol* 3D Viewer (https://www.rcsb.org/3d-view/). Threonine and Serine are the only two amino acid that were predicted to form an additional hydrogen bond between their –OH side chain (red dot) and the amide peptide bond of 150I (blue dot) on the same α chain (Figure 5B and C). IIe in DQα0402 lacks the –OH group and therefor the hydrogen bond with 150I is absent (Figure 5D). Based on the observation above, low cell surface expression of DQα0402 could be explained by the strong hydrophobicity, relatively larger side chain and the absence of hydrogen bond with 150I from the 137I substitution.

FIGURE 5.

Site-directed mutagenesis of DQα0401 AA#137 Threonine (T) residues on the DQα04:DQβ04 expression. A, Surface expression of DQ4 molecule on T2DM. B–J, Three-dimensional structures of DQβ0402:DQα0401 (B) DQβ0402:DQα0402 (C) and changes in codon #137 with serine (D) predicted by AlphaFold 3.0 (https://alphafoldserver.com/). Hydrogen bond between –OH side chain in Thr#137 or Ser#137 and NH peptide bond in Ile#150 were marked with an arrow.

DISCUSSION

In this report, we describe a single amino acid change at residue 137 of the α2 domain in DQα0402 that negatively impacts the expression of the DQ4 αβ heterodimer. By demonstrating that the same amino acid, when introduced into the DQα chain of other DQ serological groups also negatively impacted the expression of the respective DQ αβ heterodimers, we concluded that the low expression of DQ4 heterodimer was caused by the inefficient assembly of the αβ heterodimer and not due to the quantitatively low biosynthesis of the DQα0402 polypeptide chain.

Serum reaction patterns on single antigen panels are commonly utilized to define the critical amino acid(s) that is responsible for an observed serological reactivity.¹⁴ Prediction programs, such as HLA-matchmaker, assume that variability in antibody reaction patterns were because of changes in the surface-exposed amino acid structure,¹⁵ thereby explaining differential reactivity patterns. However, herein, we demonstrated that the unique DQ4 antibody reaction pattern we observed on the phenotype antigen panel was because of the level of antigen expression and not structural epitope variation as evidenced by analyzing a large number of DQ4-positive sera.

The expression level of DQβ0402 chain in the 2 DQβ0402:DQα0402 cell lines we studied (E20781 and E6452) was similar to the other 6 DQβ0402:DQα0401 cell lines despite the low abundance of DQα0402 (<5-fold) suggesting that DQβ0402 chain could efficiently form a heterodimer with a different, trans-expressed DQα chain. It has been reported that the efficiencies of DQ2/3/4 serological groups (DQA1*02/03/04/05/06 and DQB1*02/03/04) to form αβ heterodimers can vary >100-fold depending on the combinatorial pairing between different α and β chains.⁶ Although such differences effectively alter surface expression, the mechanism by which this occurs is unknown. Our data showed that a single polymorphic residue within the same serological group (ie; the same first-field of alleles) could have a significant impact on the efficiency of heterodimer formation (Figure 5B). The importance of the amino acid side chain that is capable of forming a hydrogen bond at residue α137 in the assembly/stability of the heterodimer was established through systematically substituting residue 137 of DQα0401 molecule with all 20 amino acids expressed on the T2DM transfectants. Serine was the only other amino acid that produced similar surface expression levels as the original threonine (P = 0.20212). Both serine and threonine have polar side chains with similar sizes. Most importantly, they are the only 2 amino acids that can form a hydrogen bond with the amide backbone of α150I based on the AlphaFold 3 prediction. This model suggests that α137I produced the instability of the DQαβ heterodimers by lacking the ability to form a hydrogen bond with α150I. Because of its large hydrophobic side chain, repulsion of surrounding amino acid residues cannot be ruled out. Proinflammatory cytokines such as interferon-γ can induce the expression of DQ antigens in endothelial cells.¹⁶ As such treatment was not performed in our studies, it is unclear whether cytokine treatment can overcome the instability of the heterodimers. Such a question deserves further investigation.

An even more surprising discovery was the antagonistic effect of an additional amino acid change (R23L) in the DQβ0401 chain (DQβ0401 versus DQβ0402). The DQ protein surface expression on DQβ0402:DQα0402 transfectant was approximately 5-fold lower than on the DQβ0402:DQα0401 transfectant (P = 0.00051). With the additional amino acid change in residue 23 of DQβ0401 (DQβ0401:DQα0402), the surface expression of DQ protein further reduced to half of the DQβ0402:DQα0402 expression (P = 0.00067); there was no significant difference in the expression between these 2 β chains and DQα0401 heterodimers. According to the structures predicted by AlphaFold 3, the side chain of either Arg or Leu at the position 23 of the DQβ chain is exposed on the surface of the molecule but it’s unclear how this aspect may affect the stability of the heterodimer with the DQα chain.

Donor-specific HLA antibody as a contraindication to a successful organ transplant, was first demonstrated in 1969.¹⁷ Since then, antibody reaction patterns to known HLA antigens have been used to categorize those antigens into various serological cross-reactive groups.^18,19 However, with the emergence of DNA sequencing technologies, many new HLA alleles have been identified at the molecular level but have not been characterized serologically. However, their protein sequences have been used to predict antibody reactivity by comparing shared amino acid sequence motifs with the alleles in the serologically defined cross-reactive groups.^11,14,20,21 Because an HLA antibody recognizes correctly folded tertiary structures and not linear polypeptide chain, prediction of antibody reactivity based on their structural protein sequence relies on a fundamental assumption that all related HLA polypeptides will fold and assemble into a common functional tertiary structure, exhibit similar serological reactivity and be expressed on the cell surface at similar levels.²² Current clinical practices rely on this extrapolation. In contrast to natural cellular surface expression, SABs contain a similar quantity of antigen regardless of their physiological status and might overrepresent some of the low surface expression antigens. As such, strong SAB reactivity to those low surface expressing alleles or loci may not correlate with other clinical tests (ie, physical crossmatch) or with the clinical impact. Our data showed that a single amino acid change in a location that is not previously believed to impact serological reactivity can significantly change the surface expression of the antigen. As a result, differences in reactivity patterns, as seen with the DQ4 sera used in this study, may be misinterpreted as structural differences when in fact, they represent expression-level differences. It is plausible that unappreciated low surface expression of certain HLA antigens might explain why some patients fail to show antibody mediated rejection even in the presence of high quantities of donor-specific antibody as measured by SABs as SAB beads might contain higher amounts of antigen than expressed on the cell surface.

Our discovery of structural instability of certain HLA DQ αβ heterodimers complicates the evaluation of HLA antibodies especially when considering SAB reactivity alone. SAB reactivity could overestimate the clinical impact of such antibodies thereby compromising any downstream clinical assessments. Because there are a significant number of sequence variations in exon 3 of the DQA1 and DQB1 genes that translate into amino acid polymorphisms in the α2 and β2 domains of HLA class II heterodimers, a thorough study of the surface expression level induced by such amino acid variations is warranted. If additional expression variants are identified, investigations into the clinical significance of such variants would be justified.

ACKNOWLEDGMENTS

The authors thank Dr. Afzal Nikaein for providing the leftover and deidentified serum samples.